Alexandra AI
Who it is for: University students, academic researchers, and educators in Azerbaijan and developing academic regions (e.g., BSU, ASOIU, ADA, UNEC, BMU) who must master dense academic curricula and international textbooks. What goes wrong for them today: 1. Language & Context Barrier: High-level academic literature is predominantly in English. Existing generic translation tools destroy scientific context and mathematical nuances. 2. Passive Reading & Low Retention: Students passively scroll through hundreds of PDF pages without active recall, forgetting over 70% of material within 48 hours. 3. Lack of 24/7 Personalized Socratic Mentorship: Commercial AI chatbots simply spit out final answers, robbing students of critical thinking. Students have no adaptive tutor that guides them through proofs, quizzes their understanding step-by-step, or maps out concept mastery via multidimensional radar analytics.
Preliminary score
- Value for the user14.0 / 25
A specific problem and a clear outcome from solving it.
- Prototype and use of AI14.5 / 30
A working core scenario, and what AI actually contributes to the result.
- Quality testing9.0 / 20
Tests, examples of failures, and comparison with the current approach.
- Feasibility6.5 / 15
Data requirements, running costs and a clear next step.
- Originality5.0 / 10
A meaningful difference from common solutions.
What the judges wrote
- Claude Opus 5.5Preliminary45/100
- Strengths
- Your failure write-up names specific, believable problems: Socratic leaks on proof requests, Azerbaijani terms drifting into colloquial loanwords or corrupting LaTeX JSON, and 429 rate-limit errors under concurrent quiz regeneration. Each comes with a concrete engineering response. Your focus on local universities and four-language academic fluency (AZ/EN/TR/RU) gives the tutor a clear regional angle. The README and stack show a real full-stack app, with React, an Express backend, Gemini and Firestore, rather than a slide concept.
- To improve
- You entered the Enterprise track but describe a student study tool with no business workflow or adoption metric. Either frame a buyer, such as a university department with a retention or exam-pass KPI, or say plainly why it fits. Your quantitative claims (40+ curricula tested, 82% token-cost reduction, under 5 ms latency, 70% forgetting) arrive without test tables, sample transcripts or measurement method. A small published eval would score far better, for example the number of proof prompts that leaked answers before and after the Prompt Governor. Add running-cost estimates per student, a demo video or screenshots of the core Socratic flow, and a concrete next step such as a pilot with one named course.
- GPT-6 AstraPreliminary53/100
- Strengths
- Your focus on Azerbaijani academic terminology gives the tutor a concrete local use case. Your proposed combination of Socratic guidance and curriculum-aligned quizzes makes AI relevant to the learning workflow. You also identify specific failure modes—premature solutions, terminology drift, and rate limits—rather than offering only generic success claims.
- To improve
- Your materials should demonstrate one complete textbook-to-dialogue-to-quiz session, including the resulting mastery update. Make your testing reproducible with prompts, baseline outputs, per-language results, and post-fix failure rates; distinguish cache-hit latency from fresh-generation latency and substantiate the claimed 82% cost reduction. For your chosen enterprise track, identify an institutional buyer and measurable adoption outcome, then specify content permissions, per-session operating costs, and a concrete pilot plan.
The team
Nadjwa Youssouf
business
Nicat Qurbanov
developer
Mikayıl Əskərov
developer
Huseyn HaciyevLead
product
Khayala Aliyeva
data